v1.25.2
Cloud Provider Azure v1.25.2
Changes by Kind
Bug or Regression
- ARM will possibly return StatusCode 200 and ContentLength -1 which is a bug. However, cloudprovider can handle it better by retrying. (#2315, @k8s-infra-cherrypick-robot)
- Avoid race condition between node-sync-loop routine and service reconcile routine. Avoid UpdateLoadBalancer() using stale service spec to update Load Balancers. (#2426, @k8s-infra-cherrypick-robot)
- Check and cleanup both ELB and ILB when a service is being deleted. (#2438, @jwtty)
- Fix: Skip VMSS VM update operations if the VMs are not in good state fix: Set the retryAfter to 5s when PUT failed due to concurrent operation conflict. (#2401, @k8s-infra-cherrypick-robot)
- Only check response body length when request is GET. Avoids unnecessary retry when deleting an LB. (#2390, @k8s-infra-cherrypick-robot)
Other (Cleanup or Flake)
- Chore: get node private IPs from cache but NIC (#2433, @k8s-infra-cherrypick-robot)
Dependencies
Added
Nothing has changed.
Changed
- github.com/onsi/ginkgo/v2: v2.1.6 → v2.2.0
- k8s.io/api: v0.25.0 → v0.25.2
- k8s.io/apimachinery: v0.25.0 → v0.25.2
- k8s.io/apiserver: v0.25.0 → v0.25.1-rc.0
- k8s.io/client-go: v0.25.0 → v0.25.2
- k8s.io/cloud-provider: v0.25.0 → v0.25.1-rc.0
- k8s.io/component-base: v0.25.0 → v0.25.1-rc.0
- k8s.io/component-helpers: v0.25.0 → v0.25.1-rc.0
- k8s.io/controller-manager: v0.25.0 → v0.25.1-rc.0
- k8s.io/kubelet: v0.25.0 → v0.25.1-rc.0
Removed
Nothing has changed.